Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

4404 Experiences

0 views
Skip to first unread message

forw...@csnet-sh.arpa

unread,
Jul 15, 1985, 5:24:33 PM7/15/85
to
From: obr...@CSNET-SH.ARPA

Chris Ryland asked for 4404 experiences. I had one of these
on my desk for a week a while back.

On the whole I was favorably impressed. Here are some notable
points:

o They've done some nice things to the Smalltalk system. Little
things like being able to get at the contents of a Workspace from another
workspace, certain predefined variables, etc.

o The thing runs on top UNIFLEX, a UNIX "work-alike" system.
None of the system calls are the same and none of the commands are the
same, but it still looks awfully like a UNIX.

o To hold costs down, it has a medium-resolution screen, and
hardware pan over the 1Kx1K display space. However, although there is
a Smalltalk primitive message to find out what part of the virtual space
is currently really on the screen, nobody much really uses that message
yet. Result: something can change somewhere that you can't see it, and
you won't notice that it's happened. Surprisingly, this doesn't bug
me as much as I thought it might.

o Speed: it ain't a Dorado. It's maybe 30% faster than a Dolphin,
which isn't so very terrible.

o Bugs: I found several things wrong. All seem easy to fix and
typical of a new product. None seriously hampered my test "work". I
don't know what their track record is in actually fixing any of these
since we didn't wind up buying one. The one real bogosity noted is that
when Smalltalk crashes, it makes a 2Mb core file, though there were 6 Mb
on the machine. Hence it seemed to be wasting 4Mb of core. Urk. Some
of the options in the System Workspace were therefore able to run the
system out of space and crash it.

o They fixed the Export Tracer. It now takes about 45 min. instead
of 6 hours. Also certain primitives having to do with window painting,
etc. seem much snappier than the Dolphin.

All in all they seem to know what they're doing, but this is
not a world-beating Smalltalk engine yet. Oh yes, it can also run things
like Common Lisp and Prolog as options.

Computer Sci Club

unread,
Jul 18, 1985, 11:54:54 AM7/18/85
to
In article <27...@topaz.ARPA> forw...@CSNET-SH.ARPA writes:
>From: obr...@CSNET-SH.ARPA
>
> Chris Ryland asked for 4404 experiences. I had one of these
>on my desk for a week a while back.
>
> On the whole I was favorably impressed. Here are some notable
>points:
>
> o They've done some nice things to the Smalltalk system. Little
>things like being able to get at the contents of a Workspace from another
>workspace, certain predefined variables, etc.
>
> o The thing runs on top UNIFLEX, a UNIX "work-alike" system.
>None of the system calls are the same and none of the commands are the
>same, but it still looks awfully like a UNIX.
>
> o To hold costs down, it has a medium-resolution screen, and
>hardware pan over the 1Kx1K display space. However, although there is
>a Smalltalk primitive message to find out what part of the virtual space
>is currently really on the screen, nobody much really uses that message
>yet. Result: something can change somewhere that you can't see it, and
>you won't notice that it's happened. Surprisingly, this doesn't bug
>me as much as I thought it might.

One mod I have made to the 4404 I work on is to have
StandardSystemController>controlInitialze ensure that its view is on the
screen. Thus, when you enter a new window, it scrolls into view. Secondly,
there are a number of places (PopUpMenus, FillInTheBlanks, BinaryChoices)
where checks are made to ensure that things appear completely on the
VIRTUAL screen. These should be changed to bring things up on the PHYSICAL
screen. There are no too many of these.

>
> o Speed: it ain't a Dorado. It's maybe 30% faster than a Dolphin,
>which isn't so very terrible.

As one of the few people (only?) who have used both, I must say
that I am quite impressed with the 4404. It is quite an acceptable speed,
though things like senders/implemetors should be examined for possible
speed improvements. I guess the main point is that I *dont* notice a
lack of speed.

>

-bob atkinson
U of Waterloo/ Xerox Parc

0 new messages